Skip to content
AI NEWS JOURNAL
AI, IN PERSPECTIVE

Search AI news

Search headlines, summaries and articles to connect the dots.

Newest first

Corrupting Ties: Investigating Robustness to Bias Injection Across Multi-Agent Network Topologies

arXiv:2611.21804v1 Announce Type: new Abstract: AI systems are increasingly deployed in high-stakes decision-making contexts, where interactions among components can introduce fairness risks that remain undetected when auditing components or collective outcomes in isolation. As LLM-based multi-agent systems become more prevalent, understanding how biased information propagates through agent interactions is essential for assessing their fairness. We introduce a controlled bias injection framework that replaces selected agent responses with predefined biased ones, enabling an intervention-based analysis of how network topology and injection configurations shape system behavior. We propose decision divergence metrics that quantify the effects of these interventions at the agent, network, and system levels, and evaluate five topologies on three binary classification tasks using LLMs from the Qwen and Gemma families. Sequential networks show the greatest divergence, whereas fully connected and mesh networks show less. Injection position also matters: in star networks, injecting the central agent has larger effects than injecting peripheral agents. Models further differ in whether divergence follows the injected bias. These findings highlight the role of network structure in the propagation of biased information and motivate topology-aware auditing of fairness risks in multi-agent systems.

Disambiguating Moral Circuits and Moral Features in Language Models: Moral content, normative judgment or cue words? Evidence from 14 models

arXiv:2611.21705v1 Announce Type: new Abstract: Interpretability methods can locate the attention heads and sparse autoencoder (SAE) features that change the answer a language model gives to a moral prompt, e.g., whether "Stealing medicine for a dying child is morally" continues with "justified" or "wrong". Such a component could respond to moral content, to the judgment of conduct by any norm, or to the word "morally" in the prompt. We built 227 minimal pairs of short phrases, validated by five annotators, whose two versions differ in a moral standard, in another kind of norm (prudence, etiquette, aesthetics or evidence) or in a descriptive quality, and showed them to 14 open-weight models (Llama, Gemma and Qwen, 1B to 14B, base and instruction-tuned), with and without the word "morally". In every model, the heads that change the moral answer belong to an evaluation circuit the models use for every kind of normative judgment we tested: these heads carry the same share of the effect for moral as for other normative pairs (pooled ratio 1.00, 97.5% confidence interval [0.92, 1.07], with the word; 0.90, [0.84, 0.96], without it), and the models need them as much for both. The models need these heads also when nothing in the prompt asks for a judgment, and on 400 scenarios of the ETHICS benchmark, where removing them takes away 37 to 39% of the difference between acceptable and unacceptable scenarios, against 1 to 3% for ten random heads. The SAE features we located respond mostly to the word "morally" rather than to what the phrase is about. The models represent the category of a pair (the standard the annotators assigned to it): it can be read from their hidden states, and, in the eight models we tested, removing the direction that carries it weakens the moral verdict by a few percent, and more than the other verdicts in seven of them. We release the pairs, the ratings, the code, the trained SAE dictionaries and the result files.

An Operationalization of Opaque Serial Depth

arXiv:2611.17495v1 Announce Type: new Abstract: Chain-of-thought (CoT) is currently a valuable tool for overseeing AI models, but some architectural shifts could significantly reduce CoT monitorability. We operationalize a measure that serves as a proxy for the amount of unverbalized serial cognition a model can perform, as a specific instantiation of the notion of opaque serial depth: roughly, the length of the longest path through a model's computation that does not pass through an interpretable bottleneck. Our main contribution is a particular standard for what counts as such a bottleneck. We define natural-language-rooted (NL-rooted) nodes as nodes that were initialized from a pre-training prior to produce human-understandable outputs, whose output space has not been expanded beyond that prior, and whose outputs the model has not been modified to interpret as a different type of data. By construction, the resulting measure, NLS depth, can be adjudicated before training from an architecture, training recipe, and deployment configuration alone, and is insensitive to ordinary post-training such as prompting, distillation, and RL. We find that, across open-weight models, NLS depth correlates strongly with layer count, with some exceptions, and has grown slowly over recent years. It would likely grow slowly under continued scaling of standard transformers, while latent-reasoning architectures such as full-bandwidth transformers could increase it by a large amount. We work through how NLS depth applies to a range of existing and proposed architectures, discuss alternative operationalizations, and give a worked example bounding the depth of Gemma3-1B. We argue that AI developers should track and report the NLS depth of their models as a complement to direct measurements of CoT monitorability.

Does the Ranking Hold? A Practitioner-Facing Audit of Hallucination Detectors on Small, Quantized, On-Device LLMs

arXiv:2611.17339v1 Announce Type: new Abstract: Hallucination detection for large language models is dominated by methods evaluated on 7B-70B research models with abundant sampling budgets and exact-match or ROUGE correctness labels. The small models practitioners actually ship on-device, Phi-3-mini, Llama-3.2-3B-Instruct, Qwen2.5-3B-Instruct, and Gemma-2-2B-Instruct, typically at 4-bit precision, sit outside this evidence base. We present a reproducible one-GPU audit of four detectors spanning three paradigms: length-normalized token-margin calibration, a lexical n-gram baseline, and two sampling-based consistency methods (semantic entropy and SelfCheckGPT-NLI) on two small open instruction-tuned models (Llama-3.2-3B-Instruct and Qwen2.5-3B-Instruct) loaded at 4-bit NF4 quantization on a single Colab T4 GPU, evaluated on closed-book TriviaQA. Cross-model ranking transfer is poor: Spearman rho is -0.80 and Kendall tau-b is -0.667, meaning the detector ranking actively inverts between the two models, and the cheap lexical baseline is statistically indistinguishable from the sampling-based consistency detectors on both models under bootstrap 95 percent confidence intervals. As a second, methodological contribution, we show that the standard token-overlap-to-multiple-choice correctness labeler for TruthfulQA-Generation produces near-chance AUROC (0.46-0.59) for all four detectors on both models, a labeler failure, not a detector failure, since the same detectors reach 0.69-0.89 AUROC on TriviaQA under the identical pipeline. We release the harness, prompts, and per-item scores and frame the detector choice explicitly as a fairness-of-error-rates allocation problem for practitioners deploying on resource-constrained, on-device hardware.

IWIBench: Evaluating Gradient-Free Model Driven Intra-weight Injection

arXiv:2611.08647v1 Announce Type: new Abstract: Intra-weight injection lets one model change another model's behaviour by editing its weights directly, without gradients. Such edits can range from altering lexical associations to enforcing output formats or updating factual knowledge. Despite growing interest in these approaches, there has been no systematic way to evaluate how well models perform them. We introduce IWIBench, which measures a large editor model's ability to modify the weights of a smaller target model (typically Gemma 3 4B IT) to achieve one of ten classical text-editing objectives while preserving its other capabilities. Editors are limited to four tools for reading and writing weights, with bounded tensor access and a fixed budget for generating and inspecting rollouts. A validator applies each proposed patch in a clean environment and compares the edited model against the unmodified target. Evaluating frontier closed- and open-weight models reveals sharply diverging capabilities, both across models and across tasks. A small number of models, most notably Astra and GPT-6.1 Sol, perform far ahead of the rest. The recency of this ability (little of it is evident in GPT-5.6 Sol, for instance), together with qualitative analysis, suggests it has been acquired through recent training. Tasks also vary widely: structured response resisted every attempt, and answer-length control yielded only one successful solution, whereas alias acquisition and recoding proved most amenable. Success in one task family correlated poorly with success in others, indicating an early and uneven capability frontier. We discuss strategies that editors repeatedly adopted and conclude with an outlook on intra-weight injection as an emerging but already potent approach to gradient-free behaviour modification.

BlockLock: Reproducible LLM Inference with Fixed Batch Geometry

arXiv:2611.06977v1 Announce Type: new Abstract: Reproducible LLM inference requires that a request produces the same output regardless of the other requests in the batch. Existing solutions pay for this in different ways: batch-invariant kernels constrain every operator, and isolating prefill with fixed-shape decode ({our CoRun-style baseline}) gives up shared prefill compute and pads decode. We present BlockLock, which keeps shared batched prefill and pins only the numerical path of the protected request: it fixes the geometry that the request sees, with {two modes: BlockLock-Rows uses fixed-width rows and retains batched decode of all requests; BlockLock-Pack uses terminal-anchored packing with an isolated dummy segment, then extracts the protected request's KV state for target-only decode. Under a stated per-deployment contract, both modes preserve the target's output bitwise when its companions change.} Across five {dense} models (Qwen3-8B/14B, Llama-3.1-8B-Instruct, and Gemma-3-4B/12B-IT) and three single-GPU deployments (NVIDIA A6000, A100, and H100), we evaluate our approach on both tinyBenchmarks and BFCL{; we also include initial tensor-parallel experiments on two NVIDIA A6000 GPUs. Both modes preserve target sequences and Top20-logit bits in the reported tests.} At equal delivered work, {BlockLock-Rows} reduces latency by 53.5--55.0\% and improves throughput by 115.2--122.2\% relative to vLLM Batch-Invariant; compared with batch-invariant kernels, it needs no custom kernel implementations and constrains only the protected requests.

NUMA-Torch: Compiler-Driven NUMA-Aware LLM Prefill on Multi-Socket CPUs

arXiv:2611.07105v1 Announce Type: new Abstract: For sufficiently large transformer-based Large Language Models (LLMs), whose parameters cannot be retained in on-chip caches, prefill throughput on multi-socket CPUs is often limited by sustainable DRAM bandwidth rather than peak floating-point throughput. In the standard PyTorch compilation (i.e., TorchInductor) path, model weights are allocated without considering the Non-Uniform Memory Access (NUMA) topology: specifically, the affinity between memory banks and compute cores is disregarded. This causes a substantial fraction of weight accesses to traverse the inter-socket interconnect, severely degrading the effective memory bandwidth available to each socket. We address this bottleneck through a compiler-level solution that makes TorchInductor NUMA-aware. Treating each NUMA socket as an independent tensor-parallel partition, we introduce NUMA-aware Tensor Parallelism (NUMA-TP). This approach transparently shards weights across sockets and schedules compute kernels such that thread pools and memory allocations remain strictly local to their respective sockets using CPU affinity. We evaluate the approach on GPT-2 (117M, 345M, 762M) and BERT (Small, Large), which exercise standard multi-head attention and MLP, as well as Gemma 3 270M, which combines Multi-Query Attention with a GeGLU-gated MLP. By strictly enforcing data locality, our approach successfully mitigates cross-socket traffic, achieving prefill speedups of up to 2.63x on a dual-socket Haswell platform, up to 1.87x on a dual-socket Cascade Lake platform, and up to 2.27x on a four-socket Haswell-EX platform.

Perturbation-Based Data Selection for Medical Question Answering

arXiv:2610.33440v1 Announce Type: new Abstract: Selecting a compact training subset is an important step in adapting medical language models from existing question-answering banks. We investigate perturbation-based data selection using two signals derived from Gaussian perturbations of the final-block representation: prediction-change frequency and non-flipping margin contraction. After a shared entropy prefilter, Representation-Sensitivity Prioritization (RSP) combines the signals using automatic root-mean-square scaling, while RSP-Drop ranks candidates by contraction alone. Both selectors use only questions and answer options during selection. We evaluate them with MedGemma-4B on five medical QA datasets at retained-set budgets of 64, 128, and 256 examples. At the primary budget of 128,

Timeline and related coverage

SAGE Benchmark and GAR Framework Released: Advancing Multimodal Semantic Alignment Evaluation

On October 26, 2026, arXiv published the paper arXiv:2610.31838v1, introducing Sense-Aligned Grounding Evaluation (SAGE)—a unified 10-candidate image-selection benchmark—along with the training‑agnostic Gloss-Augmented Reasoning (GAR) framework, designed for cross‑noun, verb, and entity‑centered multimodal grounding evaluation.

新闻源配图 · arXiv cs.CL

Why it matters:This study is the first to unify three types of tasks—VWSD, VerSe, and WikiDiverse—under a term-centric grounded evaluation protocol, introducing a mechanism for identifying hard negative samples and context‑exchange diagnostics. Empirical results demonstrate that the synergistic effect of definition enhancement and constraint‑guided reasoning can significantly improve the Hits@1 metric (from 65.90 to 71.22), thereby providing a more fine-grained, interpretable new evaluation paradigm for multimodal semantic alignment.

Timeline and related coverage

How Much Do You Trust A Dog? Quantifying Hint-Following in Reasoning Language Models

arXiv:2610.32055v1 Announce Type: new Abstract: Large language models (LLMs) often exhibit sycophancy, altering their answers to align with user suggestions even on objective factual and reasoning tasks. Yet the quantitative form of this influence--- and whether it is calibrated to source reliability--- remains poorly understood. In this work, we investigate this behavior through hint-induced answer shifts on MMLU questions, evaluating prompts of the form ``[Source] thinks the answer is (X)''. We systematically vary the source to which the hint is attributed (a professor, a friend, or a dog) across three open-source model families, Qwen, Gemma, and SmolLM. We show that the change induced by the hint follows a predictable structure well-characterized by an exponential tilt: the hint induces

Timeline and related coverage

Research reveals the local geometric structure of semantic meaning in large language models

A paper published on October 24, 2026, on arXiv (ID 2610.30715v1) characterizes the local geometric properties of semantic meaning in embedding space across four decoder-only large language models—Gemma 4, Llama 3.2, Qwen 2.5, and Pythia—using cross-lingual probes and fine-grained probes.

Meaning Contracts: Local Semantic Geometry in Large Language Models

Why it matters:This study is the first to systematically characterize the invariance, variability, and refinement mechanisms within “the same semantic factor” in LLMs, proposing the “local anisotropic neighborhood” model. The approach has been validated through blind human evaluations, demonstrating rigorous methodology, broad model coverage, and cross-architecture consistency in its findings.

Timeline and related coverage

SIM-UT: A User Behavior Simulation and Evaluation Framework for Medical Triage Chatbots

Researchers have proposed SIM-UT (Simulation of User Behaviour for Triage), a multi-round, real-user-behavior–based evaluation method designed to identify eco‑valid failure patterns of large language model (LLM)-driven medical triage chatbots in real-world settings within low- and middle-income countries (LMICs).

SIM-UT: A User Simulation based Evaluation Framework for Medical Triage Chatbots

Why it matters:This framework breaks through the limitations of existing single-round, English‑centric evaluations by conducting over 5,000 rounds of multilingual simulated conversations, covering real‑world WhatsApp health programs in India, South Africa, and Kenya. For the first time, it systematically reveals “de‑escalation risk,” a critical failure mode, demonstrating both methodological innovation and significant clinical safety implications.

Timeline and related coverage